How readers attentive and inattentive to task-related information scan: An eye-tracking study

Abstract

Purpose

Attentive to task-related information should be the highest priority for all readers engaged in task-reading. Investigating the scanning behaviors of attentive versus inattentive readers shed new insights into the sequential cognitive processes, but it has seldom been studied. This study investigates their global patterns on a global scale, pertaining to the whole length of scanpaths, and further compares local tactics, local strategies, and local strategy transitions on a local scale, related to the isolated regions of scanpath.

Design/methodology/approach

A regular style reading system with the question, navigating, and text areas on its interface, and two types of task, namely fact-finding (FF) and content understanding (CU), were designed in an eye-tracking experiment. 24 participants were placed into attentive (AR) or inattentive (IAR) readers groups according to their fixation duration on task-related paragraphs. A global sequence analysis algorithm, Needleman-Wunsch, was applied to uncover global patterns across the whole length of scanpaths (whole-scanpaths). A local sequence analysis method related to frequent sub-scanpaths was adopted to extract local tactics specific to the reader and task. Coding was performed to identify local strategies by classifying local tactics. A local strategy transition was further identified as a sequence of frequent local strategies at the beginning, middle, and ending phases.

Findings

Whole-scanpaths of AR significantly differed from those of IAR, despite the absence of global patterns for each group. Five types of local strategy were identified, namely locating information (LI), evaluating and verifying text relevance (EVR), navigation heuristics (NH), synthesizing information (SI), and contextual clues (CC). AR applied all types in both tasks, whereas IAR applied only two types and stuck with EVR. Furthermore, two types of local strategy transition were identified: comprehensive exploration and iterative content evaluation. AR employed the former with the linear feature in FF and the spiral feature in CU, while IAR employed the latter in both tasks.

Originality

This study advances the knowledge of dynamic cognitive processing from an attentive and inattentive to task-related information perspective. An objective analysis perspective for obtaining global patterns, local tactics, local strategies, and local strategy transitions is provided, then it can provide new insights into automatically classifying readers. The results also generate detailed and valuable guidance for improving reading system design and training readers.